Dissecting the biophysics and biology of intrinsically disordered proteins.

Intrinsically disordered regions (IDRs) within human proteins play critical roles in cellular information processing, including signaling, transcription, stress response, DNA repair, genome organization, and RNA processing. Here, we summarize current challenges in the field and propose cutting-edge approaches to address them in physiology and disease processes, with a focus on cancer.

Folding-upon-binding pathways of an intrinsically disordered protein from a deep Markov state model.

A central challenge in the study of intrinsically disordered proteins is the characterization of the mechanisms by which they bind their physiological interaction partners. Here, we utilize a deep learning-based Markov state modeling approach to characterize the folding-upon-binding pathways observed in a long timescale molecular dynamics simulation of a disordered region of the measles virus nucleoprotein NTAIL reversibly binding the X domain of the measles virus phosphoprotein complex. We find that folding-upon-binding predominantly occurs via two distinct encounter complexes that are differentiated by the binding orientation, helical content, and conformational heterogeneity of NTAIL. We observe that folding-upon-binding predominantly proceeds through a multi-step induced fit mechanism with several intermediates and do not find evidence for the existence of canonical conformational selection pathways. We observe four kinetically separated native-like bound states that interconvert on timescales of eighty to five hundred nanoseconds. These bound states share a core set of native intermolecular contacts and stable NTAIL helices and are differentiated by a sequential formation of native and non-native contacts and additional helical turns. Our analyses provide an atomic resolution structural description of intermediate states in a folding-upon-binding pathway and elucidate the nature of the kinetic barriers between metastable states in a dynamic and heterogenous, or "fuzzy", protein complex.

Molecular Basis of Small-Molecule Binding to α-Synuclein.

Intrinsically disordered proteins (IDPs) are implicated in many human diseases. They have generally not been amenable to conventional structure-based drug design, however, because their intrinsic conformational variability has precluded an atomic-level understanding of their binding to small molecules. Here we present long-time-scale, atomic-level molecular dynamics (MD) simulations of monomeric α-synuclein (an IDP whose aggregation is associated with Parkinson's disease) binding the small-molecule drug fasudil in which the observed protein-ligand interactions were found to be in good agreement with previously reported NMR chemical shift data. In our simulations, fasudil, when bound, favored certain charge-charge and π-stacking interactions near the C terminus of α-synuclein but tended not to form these interactions simultaneously, rather breaking one of these interactions and forming another nearby (a mechanism we term dynamic shuttling). Further simulations with small molecules chosen to modify these interactions yielded binding affinities and key structural features of binding consistent with subsequent NMR experiments, suggesting the potential for MD-based strategies to facilitate the rational design of small molecules that bind with disordered proteins.

Developing a molecular dynamics force field for both folded and disordered protein states.

Molecular dynamics (MD) simulation is a valuable tool for characterizing the structural dynamics of folded proteins and should be similarly applicable to disordered proteins and proteins with both folded and disordered regions. It has been unclear, however, whether any physical model (force field) used in MD simulations accurately describes both folded and disordered proteins. Here, we select a benchmark set of 21 systems, including folded and disordered proteins, simulate these systems with six state-of-the-art force fields, and compare the results to over 9,000 available experimental data points. We find that none of the tested force fields simultaneously provided accurate descriptions of folded proteins, of the dimensions of disordered proteins, and of the secondary structure propensities of disordered proteins. Guided by simulation results on a subset of our benchmark, however, we modified parameters of one force field, achieving excellent agreement with experiment for disordered proteins, while maintaining state-of-the-art accuracy for folded proteins. The resulting force field, a99SB-disp, should thus greatly expand the range of biological systems amenable to MD simulation. A similar approach could be taken to improve other force fields.

Quantifying the Relationship between Conformational Dynamics and Enzymatic Activity in Ribonuclease HI Homologues.

Ribonuclease HI (RNHI), a ubiquitous, non-sequence-specific endonuclease, cleaves the RNA strand in RNA/DNA hybrids. RNHI functions in replication and genome maintenance, and retroviral reverse transcriptases contain an essential ribonuclease H domain. Nuclear magnetic resonance (NMR) spectroscopy combined with molecular dynamics (MD) simulations suggests a model in which the extended handle region domain of Escherichia coli RNHI populates (substrate-binding-competent) "open" and (substrate-binding-incompetent) "closed" states, while the thermophilic Thermus thermophilus RNHI mainly populates the closed state at 300 K [Stafford, K. A., Robustelli, P., and Palmer, A. G., III (2013) PLoS Comput. Biol. 9, 1-10]. In addition, an in silico-designed mutant E. coli Val98Ala RNHI was predicted to populate primarily the closed state. The work presented here validates this model and confirms the predicted properties of the designed mutant. MD simulations suggest that the conformational preferences of the handle region correlate with the conformations of Trp85, Thr92, and Val101. NMR residual dipolar coupling constants, three-bond scalar coupling constants, and chemical shifts experimentally define the conformational states of these residues and hence of the handle domain. These NMR parameters correlate with the Michaelis constants for RNHI homologues, confirming the important role of the handle region in the modulation of substrate recognition and illustrating the power of NMR spectroscopy in dissecting the conformational preferences underlying enzyme function.

Mechanism of Coupled Folding-upon-Binding of an Intrinsically Disordered Protein.

Intrinsically disordered proteins (IDPs), which in isolation do not adopt a well-defined tertiary structure but instead populate a structurally heterogeneous ensemble of interconverting states, play important roles in many biological pathways. IDPs often fold into ordered states upon binding to their physiological interaction partners (a so-called "folding-upon-binding" process), but it has proven difficult to obtain an atomic-level description of the structural mechanisms by which they do so. Here, we describe in atomic detail the folding-upon-binding mechanism of an IDP segment to its binding partner, as observed in unbiased molecular dynamics simulations. In our simulations, we observed over 70 binding and unbinding events between the α-helical molecular recognition element (α-MoRE) of the intrinsically disordered C-terminal domain of the measles virus nucleoprotein (NTAIL) and the X domain (XD) of the measles virus phosphoprotein complex. We found that folding-upon-binding primarily occurred through induced-folding pathways (in which intermolecular contacts form before or concurrently with the secondary structure of the disordered protein)-an observation supported by previous experiments-and that the transition state ensemble was characterized by formation of just a few key intermolecular contacts and was otherwise highly structurally heterogeneous. We found that when a large amount of helical content was present early in a transition path, NTAIL typically unfolded and then refolded after additional intermolecular contacts formed. We also found that, among conformations with similar numbers of intermolecular contacts, those with less helical content had a higher probability of ultimately forming the native complex than conformations with more helical content, which were more likely to unbind. These observations suggest that even after intermolecular contacts have formed, disordered regions can have a kinetic advantage over folded regions in the folding-upon-binding process.

Thermal adaptation of conformational dynamics in ribonuclease H.

The relationship between inherent internal conformational processes and enzymatic activity or thermodynamic stability of proteins has proven difficult to characterize. The study of homologous proteins with differing thermostabilities offers an especially useful approach for understanding the functional aspects of conformational dynamics. In particular, ribonuclease HI (RNase H), an 18 kD globular protein that hydrolyzes the RNA strand of RNA:DNA hybrid substrates, has been extensively studied by NMR spectroscopy to characterize the differences in dynamics between homologs from the mesophilic organism E. coli and the thermophilic organism T. thermophilus. Herein, molecular dynamics simulations are reported for five homologous RNase H proteins of varying thermostabilities and enzymatic activities from organisms of markedly different preferred growth temperatures. For the E. coli and T. thermophilus proteins, strong agreement is obtained between simulated and experimental values for NMR order parameters and for dynamically averaged chemical shifts, suggesting that these simulations can be a productive platform for predicting the effects of individual amino acid residues on dynamic behavior. Analyses of the simulations reveal that a single residue differentiates between two different and otherwise conserved dynamic processes in a region of the protein known to form part of the substrate-binding interface. Additional key residues within these two categories are identified through the temperature-dependence of these conformational processes.

Folding-upon-binding pathways of an intrinsically disordered protein from a deep Markov state model.

A central challenge in the study of intrinsically disordered proteins is the characterization of the mechanisms by which they bind their physiological interaction partners. Here, we utilize a deep learning based Markov state modeling approach to characterize the folding-upon-binding pathways observed in a long-time scale molecular dynamics simulation of a disordered region of the measles virus nucleoprotein NTAIL reversibly binding the X domain of the measles virus phosphoprotein complex. We find that folding-upon-binding predominantly occurs via two distinct encounter complexes that are differentiated by the binding orientation, helical content, and conformational heterogeneity of NTAIL. We do not, however, find evidence for the existence of canonical conformational selection or induced fit binding pathways. We observe four kinetically separated native-like bound states that interconvert on time scales of eighty to five hundred nanoseconds. These bound states share a core set of native intermolecular contacts and stable NTAIL helices and are differentiated by a sequential formation of native and non-native contacts and additional helical turns. Our analyses provide an atomic resolution structural description of intermediate states in a folding-upon-binding pathway and elucidate the nature of the kinetic barriers between metastable states in a dynamic and heterogenous, or "fuzzy", protein complex.

Clustering Heterogeneous Conformational Ensembles of Intrinsically Disordered Proteins with t-Distributed Stochastic Neighbor Embedding.

Intrinsically disordered proteins (IDPs) populate a range of conformations that are best described by a heterogeneous ensemble. Grouping an IDP ensemble into "structurally similar" clusters for visualization, interpretation, and analysis purposes is a much-desired but formidable task, as the conformational space of IDPs is inherently high-dimensional and reduction techniques often result in ambiguous classifications. Here, we employ the t-distributed stochastic neighbor embedding (t-SNE) technique to generate homogeneous clusters of IDP conformations from the full heterogeneous ensemble. We illustrate the utility of t-SNE by clustering conformations of two disordered proteins, Aß42, and α-synuclein, in their APO states and when bound to small molecule ligands. Our results shed light on ordered substates within disordered ensembles and provide structural and mechanistic insights into binding modes that confer specificity and affinity in IDP ligand binding. t-SNE projections preserve the local neighborhood information, provide interpretable visualizations of the conformational heterogeneity within each ensemble, and enable the quantification of cluster populations and their relative shifts upon ligand binding. Our approach provides a new framework for detailed investigations of the thermodynamics and kinetics of IDP ligand binding and will aid rational drug design for IDPs.

Rational optimization of a transcription factor activation domain inhibitor.

Transcription factors are among the most attractive therapeutic targets but are considered largely 'undruggable' in part due to the intrinsically disordered nature of their activation domains. Here we show that the aromatic character of the activation domain of the androgen receptor, a therapeutic target for castration-resistant prostate cancer, is key for its activity as transcription factor, allowing it to translocate to the nucleus and partition into transcriptional condensates upon activation by androgens. On the basis of our understanding of the interactions stabilizing such condensates and of the structure that the domain adopts upon condensation, we optimized the structure of a small-molecule inhibitor previously identified by phenotypic screening. The optimized compounds had more affinity for their target, inhibited androgen-receptor-dependent transcriptional programs, and had an antitumorigenic effect in models of castration-resistant prostate cancer in cells and in vivo. These results suggest that it is possible to rationally optimize, and potentially even to design, small molecules that target the activation domains of oncogenic transcription factors.

Interpreting protein structural dynamics from NMR chemical shifts.

In this investigation, semiempirical NMR chemical shift prediction methods are used to evaluate the dynamically averaged values of backbone chemical shifts obtained from unbiased molecular dynamics (MD) simulations of proteins. MD-averaged chemical shift predictions generally improve agreement with experimental values when compared to predictions made from static X-ray structures. Improved chemical shift predictions result from population-weighted sampling of multiple conformational states and from sampling smaller fluctuations within conformational basins. Improved chemical shift predictions also result from discrete changes to conformations observed in X-ray structures, which may result from crystal contacts, and are not always reflective of conformational dynamics in solution. Chemical shifts are sensitive reporters of fluctuations in backbone and side chain torsional angles, and averaged (1)H chemical shifts are particularly sensitive reporters of fluctuations in aromatic ring positions and geometries of hydrogen bonds. In addition, poor predictions of MD-averaged chemical shifts can identify spurious conformations and motions observed in MD simulations that may result from force field deficiencies or insufficient sampling and can also suggest subsets of conformational space that are more consistent with experimental data. These results suggest that the analysis of dynamically averaged NMR chemical shifts from MD simulations can serve as a powerful approach for characterizing protein motions in atomistic detail.

Characterization of the conformational equilibrium between the two major substates of RNase A using NMR chemical shifts.

Following the recognition that NMR chemical shifts can be used for protein structure determination, rapid advances have recently been made in methods for extending this strategy for proteins and protein complexes of increasing size and complexity. A remaining major challenge is to develop approaches to exploit the information contained in the chemical shifts about conformational fluctuations in native states of proteins. In this work we show that it is possible to determine an ensemble of conformations representing the free energy surface of RNase A using chemical shifts as replica-averaged restraints in molecular dynamics simulations. Analysis of this surface indicates that chemical shifts can be used to characterize the conformational equilibrium between the two major substates of this protein.

Small molecules targeting the disordered transactivation domain of the androgen receptor induce the formation of collapsed helical states.

Intrinsically disordered proteins, which do not adopt well-defined structures under physiological conditions, are implicated in many human diseases. Small molecules that target the disordered transactivation domain of the androgen receptor have entered human trials for the treatment of castration-resistant prostate cancer (CRPC), but no structural or mechanistic rationale exists to explain their inhibition mechanisms or relative potencies. Here, we utilize all-atom molecular dynamics computer simulations to elucidate atomically detailed binding mechanisms of the compounds EPI-002 and EPI-7170 to the androgen receptor. Our simulations reveal that both compounds bind at the interface of two transiently helical regions and induce the formation of partially folded collapsed helical states. We find that EPI-7170 binds androgen receptor more tightly than EPI-002 and we identify a network of intermolecular interactions that drives higher affinity binding. Our results suggest strategies for developing more potent androgen receptor inhibitors and general strategies for disordered protein drug design.

Determination of protein structures in the solid state from NMR chemical shifts.

Solid-state NMR spectroscopy does not require proteins to form crystalline or soluble samples and can thus be applied under a variety of conditions, including precipitates, gels, and microcrystals. It has recently been shown that NMR chemical shifts can be used to determine the structures of the native states of proteins in solution. By considering the cases of two proteins, GB1 and SH3, we provide an initial demonstration here that this type of approach can be extended to the use of solid-state NMR chemical shifts to obtain protein structures in the solid state without the need for measuring interatomic distances.

Development of a Force Field for the Simulation of Single-Chain Proteins and Protein-Protein Complexes.

The accuracy of atomistic physics-based force fields for the simulation of biological macromolecules has typically been benchmarked experimentally using biophysical data from simple, often single-chain systems. In the case of proteins, the careful refinement of force field parameters associated with torsion-angle potentials and the use of improved water models have enabled a great deal of progress toward the highly accurate simulation of such monomeric systems in both folded and, more recently, disordered states. In living organisms, however, proteins constantly interact with other macromolecules, such as proteins and nucleic acids, and these interactions are often essential for proper biological function. Here, we show that state-of-the-art force fields tuned to provide an accurate description of both ordered and disordered proteins can be limited in their ability to accurately describe protein-protein complexes. This observation prompted us to perform an extensive reparameterization of one variant of the Amber protein force field. Our objective involved refitting not only the parameters associated with torsion-angle potentials but also the parameters used to model nonbonded interactions, the specification of which is expected to be central to the accurate description of multicomponent systems. The resulting force field, which we call DES-Amber, allows for more accurate simulations of protein-protein complexes, while still providing a state-of-the-art description of both ordered and disordered single-chain proteins. Despite the improvements, calculated protein-protein association free energies still appear to deviate substantially from experiment, a result suggesting that more fundamental changes to the force field, such as the explicit treatment of polarization effects, may simultaneously further improve the modeling of single-chain proteins and protein-protein complexes.

Fast and accurate predictions of protein NMR chemical shifts from interatomic distances.

We present a method, CamShift, for the rapid and accurate prediction of NMR chemical shifts from protein structures. The calculations performed by CamShift are based on an approximate expression of the chemical shifts in terms of polynomial functions of interatomic distances. Since these functions are very fast to compute and readily differentiable, the CamShift approach can be utilized in standard protein structure calculation protocols.

Folding of small proteins by Monte Carlo simulations with chemical shift restraints without the use of molecular fragment replacement or structural homology.

It has recently been shown that protein structures can be determined from nuclear magnetic resonance (NMR) chemical shifts using a molecular fragment replacement strategy. In these approaches, structural motifs are selected from existing protein structures on the basis of chemical shift and sequence homology and assembled to generate new structures. Here, we demonstrate that it is also possible to determine structures of proteins by directly incorporating experimental NMR chemical shifts as structural restraints in conformational searches, without the use of structural homology and molecular fragment replacement. In this approach, a protein is folded from an extended conformation to its native state using a simulated annealing procedure that minimizes an energy function that combines a standard force field with a term that penalizes the differences between experimental and calculated chemical shifts. We provide an initial demonstration of this procedure by determining the structure of two small proteins, with alpha and beta folds, respectively.

Molecular switch based on a biologically important redox reaction.

Building on our earlier report of a single-molecule probe, we show how biologically important redox centers, nicotinamide and quinone, incorporated into a fluorophore-spacer-receptor molecular structure, form redox active molecular switches, with the photoinduced electron-transfer behavior of each depending on the oxidation state of the receptor subunit. The switch based on nicotinamide (3/6) is strongly fluorescent in its oxidized state (Phi(F) approximately 1.0) but nonfluorescent in the reduced state (Phi(F) < 0.001) due to electron transfer from the reduced nicotinamide to the photoexcited fluorophore. The fluorescence can be reversibly switched off and on chemically by successive reduction with NaBH(3)CN and oxidation with tetrachlorobenzoquinone and switched electrochemically over 10 cycles without significant degradation. A similar switch based on quinonimine turned out to be nonfluorescent in both reduced and oxidized states: in addition to a similar quenching mechanism in the reduced state, quenching also occurs in the oxidized state, due to electron transfer from the fluorophore to the receptor. Ab initio quantum chemical calculations of orbital energy levels were used to corroborate these quenching mechanisms. Calculations predicted photoinduced electron transfer to be energetically favorable in all cases where quenching was observed and unfavorable in all cases where it was not. Application of the perylene analogue as a biosensor has also been demonstrated by coupling the switch to the catalytic pathway of yeast alcohol dehydrogenase, a common NADH/NAD(+)-utilizing enzyme.

Water dispersion interactions strongly influence simulated structural properties of disordered protein states.

Many proteins can be partially or completely disordered under physiological conditions. Structural characterization of these disordered states using experimental methods can be challenging, since they are composed of a structurally heterogeneous ensemble of conformations rather than a single dominant conformation. Molecular dynamics (MD) simulations should in principle provide an ideal tool for elucidating the composition and behavior of disordered states at an atomic level of detail. Unfortunately, MD simulations using current physics-based models tend to produce disordered-state ensembles that are structurally too compact relative to experiments. We find that the water models typically used in MD simulations significantly underestimate London dispersion interactions, and speculate that this may be a possible reason for these erroneous results. To test this hypothesis, we create a new water model, TIP4P-D, that approximately corrects for these deficiencies in modeling water dispersion interactions while maintaining compatibility with existing physics-based models. We show that simulations of solvated proteins using this new water model typically result in disordered states that are substantially more expanded and in better agreement with experiment. These results represent a significant step toward extending the range of applicability of MD simulations to include the study of (partially or fully) disordered protein states.

Functional implications of large backbone amplitude motions of the glycoprotein 130-binding epitope of interleukin-6.

Human interleukin (IL)-6 plays a pivotal role in the immune response, hematopoiesis, the acute-phase response, and inflammation. IL-6 has three distinct receptor epitopes, termed sites I, II, and III, that facilitate the formation of a signaling complex. IL-6 signals via a homodimer of glycoprotein 130 (gp130) after initially forming a heterodimer with the nonsignaling α-receptor [IL-6 α-receptor (IL-6R)] via site I. Here, we present the backbone dynamics of apo-IL-6 as determined by analysis of NMR relaxation data with the extended model-free formalism of Lipari and Szabo. To alleviate significant resonance overlap in the HSQC-type spectra, cell-free protein synthesis was used to selectively (15) N-label residues, thereby ensuring a complete set of residue-specific dynamics. The calculated order parameters [square of the generalized model-free order parameter (S(2))] showed significant conformational heterogeneity among clusters of residues in IL-6. In particular, the N-terminal region of the long AB-loop, which corresponds spatially to one of the gp130 receptor binding epitopes (i.e. site III), experiences substantial fluctuations along the conformation of the main chain (S(2) = 0.3-0.8) that are not observed at the other two epitopes or in other cytokines. Thus, we postulate that dynamic properties of the AB-loop are responsible for inhibiting the interaction of IL-6 with gp130 in the absence of the IL-6R, and that binding of IL-6R at site I shifts the dynamic equilibrium to favor interaction with gp130 at site III. In addition, molecular dynamics simulations corroborated the NMR-derived dynamics, and showed that the BC-loop adopts different substates that possibly play a role in facilitating receptor assembly.